Development of a low cost data acquisition system and training of J48 algorithm for classifying faults in cutting tool

Abstract

The personalized use of the single-point cutting tool plays an crucial role in the economy of machining operations. Eventually, any machining defect in the cutting tool leads to the deterioration of the complete machining activity. Such faults are not traceable to conventional condition monitoring practices. The characterization of such in-process tooling defects must be approached intelligently. This would also contribute to the “self-monitoring” requirement in Industry 4.0. In this regard, the following is an introduction to a supervised machine learning (ML) classifier for designing empirical classification models for monitoring of tool condition. During the process of turning performed on the lathe, the vibration responses in the time domain of various defective and sound configurations of the single cutting tool were collected. By designing an algorithm which is event-based in Python programming, additional statistical features were extracted. J48 algorithm was used to display the features in the decision tree (DT) for developing data acquisition systems (DAQ). The DT Classifier affords accuracy of as much as 95 % and resulted to be a satisfactory classifier algorithm. The approach of implementation is maximum appropriate for actual time implementation of fault diagnosis. The system is small, low cost and additionally portable.